RunLoyal MCP server
# RunLoyal MCP server for ChatGPT and Claude
RunLoyal publishes Partner APIs for the pet owner booking flow. Cotlet.dev connects ChatGPT, Claude and other AI agents to your signed-in RunLoyal Business Portal with 40 hosted tools for pet counts, bookings, report cards, messages and reports. Write tools ask for confirmation by default. Cotlet.dev is independent and is not affiliated with RunLoyal.
## What you can ask
> Read tomorrow's pet count and available boarding rooms.
> Draft today's report cards and ask before sending them.
> Quote a boarding stay before booking it.
> Review vaccinations that need attention.
> Read open invoices and membership reports.
## Connect in 2 minutes
Get your API key at https://cotlet.dev/app and export it as `COTLET_API_KEY`.
### ChatGPT
Settings → Apps → add connector URL `https://cotlet.dev/mcp/v1/connect`, then sign in with Cotlet.dev. No API key needed.
### Claude
Settings > Connectors > Add custom connector: `https://cotlet.dev/mcp/v1/connect`, then sign in to Cotlet.dev.
### Claude Code
```bash
claude mcp add --transport http cotlet https://cotlet.dev/mcp/v1 --header "Authorization: Bearer $COTLET_API_KEY"
```
### Cursor
Copy `.cursor/mcp.json` from this repo into your project:
```json
{
"mcpServers": {
"cotlet": {
"url": "https://cotlet.dev/mcp/v1",
"headers": {
"Authorization": "Bearer ${env:COTLET_API_KEY}"
}
}
}
}
```
## Run the local MCP bridge
Python 3.10+, with no dependencies. Clone this repository, set your Cotlet.dev API key, then run:
```bash
export COTLET_API_KEY="your-key"
python3 runloyal_agent.py serve
```
This local stdio bridge lists seven browser helper tools: `runloyal_guide`, `runloyal_status`, `runloyal_open`, `runloyal_read`, `runloyal_click`, `runloyal_type` and `runloyal_close`.
It forwards calls to the hosted Cotlet.dev MCP server.
The 40 task tools described on the integration page are available through the hosted connection.
Without a key, the local bridge can list its helpers; calls return setup instructions.
```bash
python3 runloyal_agent.py connect
python3 runloyal_agent.py guide
python3 runloyal_agent.py status
```
## Questions from RunLoyal users
**Does RunLoyal have an API or MCP server?**
RunLoyal publishes Partner API documentation at api.runloyal.com for the pet owner booking flow. The Cotlet.dev page, checked 17 Nov 2026, found no official RunLoyal MCP server or ChatGPT or Claude connector.
**How is this different from Scout AI?**
Scout AI works inside RunLoyal on report card notes, messages, photos and scheduling trends. Cotlet.dev lets ChatGPT and Claude work in your signed-in Business Portal.
**Do add-on features work?**
Only when your RunLoyal account includes them. Memberships, Dynamic Pricing, Vax Verify, Route Optimizer and SMS need their respective add-ons.
**Which RunLoyal user should I use?**
Sign in with a staff user whose staff group has the sections, actions and reports the agent should use.
## Integration details
[RunLoyal tools and setup on Cotlet.dev](https://cotlet.dev/integrations/runloyal-api-mcp-server?utm_source=github&utm_medium=readme&utm_campaign=runloyal)
Start free: https://cotlet.dev/signup?app=runloyal&utm_source=github&utm_medium=readme&utm_campaign=runloyal
TDQS
Scored across 7 tools
Each tool maps to a distinct phase of work: guide (docs), status (account/credit readiness), open (navigation), read (page inspection), click/type (interaction), close (teardown). The numbered-element model in read/click/type makes their boundary explicit and non-overlapping. Guide vs. status are the closest pair but differ clearly (domain how-to vs. connection/credits health).
All seven tools share the runloyal_ prefix in snake_case, which is highly predictable. Minor deviation: 'guide' and 'status' are nouns while the rest are imperative verbs (open, read, click, type, close).
Seven tools is a tight, well-scoped surface for a browser-automation wrapper — one setup/diagnostic pair, one navigation pair, two interaction primitives, and a cleanup tool. No redundant or speculative tools.
Open/read/click/type/close plus read/status covers the core browser loop and lets agents avoid dead ends on most flows. Gaps remain for common interactions: no keyboard/Enter or form-submit primitive, no dropdown-select, and no scroll or wait for long or asynchronously loading pages.